MMSE RECONSTRUCTION FOR 3D ULTRASOUND IMAGES (TA-P7)

Author(s) : Wei Huang (University of Virginia, USA) Yibin Zheng (University of Virginia, USA) Janelle A. Molloy (University of Virginia, USA)

Abstract : The reconstruction of 3D ultrasound (US) images from a set of irregularly spaced B-scans is of great interest in clinic procedures. Due to the presence of speckle, a form of multiplicative noise, 3D US reconstruction for high image quality data is a difficult task. Conventional reconstruction algorithms have low computational com-plexity but the reconstructed volume suffers from severe speckle contamination. In this paper, we propose a novel reconstruction method, which is optimal in minimum mean-squared error (MMSE) sense with a multiplicative noise model. Simulation results for synthetic US images, as well as experimental data are presented.

Menu